Date of this version: Jun 22 2021
A reviewer of a project of ours on colour learning says the following:
The critical problem of this study is the presence of covariation between (at least) three features of the object stimulus, its colour, its size, and its shape. Therefore, it is not possible to conclude whether the observed difference in the learned preference between the two stimuli was due to the colour or to the shape or something else.
In order to address this point we are re-doing the study with this critique in mind. We are conducting a study where the goal is to determine whether guppies can change their preference for a coloured object, regardless of its colour. Guppies are trained to green and blue objects that differ only in their colours and not also their sizes and shapes (Figure 1).
In this overview steps 1 - 3 are largely similar to experiment 1. This time the experiment is done with a large variety of objects that only differ in their colour. Steps 4 - 6 are additional trials for this second experiment so that, in addition to making sure learning is specific to colour, we also see how robust learning to colour is across different objects of similar colours.
During training, on any given day all guppies will get reinforced for visiting the same object pair. The presentation of the four training object pairs will have a randomized order that is the same for all guppies.
Objects are cleaned by rinsing with water conditioned with SeaChem Prime between trials. For example, if a single training object was used, a scratch could mean that particular object was treated differently from other objects, so we would be looking at learning about a scratch, not colour.
Generalization 1 refers to a generalization to different colours of Lego material whereas Generalization 2 refers to generalization to completely different objects (green and blue disks).
Figure 1: The 7 pairs of objects for the experiment as well as an experimental timeline. The numbered object pairs are used for training while the lettered object pairs are used for testing. The experimental timeline represents the order of the trials. The open orange square at trial 0 represents the initial test which is done with object pair A. The filled green circles represent the 20 training trials (trials 1-20). The filled orange square is the final test trial which is done with object pair A. The 2 red circles represent the refresher trials which are done with object pairs 2 and 3 (I picked these with a random number generator between 1 and 4). The open cyan triangle represents generalization 1 which is done with object pair B. The open blue circle is generalization 2 which is done with object pair C
| Object Colour | Colour Family | Exact Colour | Hex Value |
|---|---|---|---|
| Green | Dark Green | Bright Yellowish Green | B5BD00 |
| Blue | Blue | Dark Azur |
The objects were 6 pairs of Lego blocks with the manufacturer colours A and B. The specific models for the object pairs are as follows
| Object Pair | Brick Style | Design ID |
|---|---|---|
| Pair A | BRICK 1X2 | 3004 |
| Pair B | BRICK | |
| Pair 1 | BRICK 1X2 | 3004 |
| Pair 2 | BRICK 2X4 | 3001 |
| Pair 3 | BRICK 2X2 | 30165 |
| Pair 4 | BRICK 1X1 | 3005 |
These can be looked up and/or purchased using the official Lego store’s Choose a brick service by entering the design ID in the Enter an item number or design ID prompt of the Enter set number section.
In this section we provide step by step instructions on how to perform an experiment for the day.
Briefly the protocol involves the following: First you need to make the food strips and then wait one hour for them to fix onto the strips. Then you run the trials with the fish. Each trial lasts 5 minutes. Guppies are tested and trained in the same order every day.
Throughout all trials the side of the tank the rewarding object is on should be switched. I have written an R script that has produced a data sheet which randomly assigned individuals to treatments as well as which sides of the tank should be rewarded during training for each guppy.
The script ensures on any given day the position of the object is random but across all trials the number of left reinforced and right reinforced trials is even across guppies on any given day as well within an individual guppy across all days. Consult this sheet when performing the experiment to know which objects go where for which fish on any particular day.
After performing the trials for the day you transfer the video files from the SD
card on the camera and into the C://Users/Reader Lab/2-colour-learning-experiment-2/videos/
folder on the desktop. Finally, you use EthoVision to extract the
positional data from the video files.
Some additional details. Guppies should only be fed Tetramin flake food in their home tank every three days during this experiment. This is to increase the motivation to learn and perform the task.
The first thing to do during a training day is to make food strips. 16 strips are prepared, one for each guppy. The food strips that are attached to the objects are made with the following protocol:
Before you start performing the trials the SD card needs to have footage from the previous day cleared up.
Computer click Removable Disk (L:) 3, Double click MP_ROOT then
100ANV01Date modified date that corresponds
to the previous dayCTRL while selecting the files that have a type of MP4 VideoComputer click Removable Disk (L:)PRIVATE/AVCHD/BDMV/STREAM/Date modified date that corresponds
to the previous dayCTRL while selecting the files that have a type of MP4 VideoDeleteComputer right click Removable Disk (L:) and select ejectIf the current trial is a training trial you do the following:
If the trial is a test trial you do the above EXCEPT there is no need to add food to a plastic strip. Instead use never before used blank plastic strips that I provide. On test trial days you therefore do not need to wait an extra hour.
L://MP_ROOT/.mp4 video files from the trials of the current day into a
folder named after the trial number (e.g., trial-x/) found within the
C://Users/Reader Lab/2-colour-learning-experiment-2/videos/ folder..mp4 files based on the IDs of the fish. Each video file should
be named after a specific fish.New from templateUse a custom templateBrowse... and navigate to C://Users/Reader Lab/2-colour-learning-experiment-2/trial-2/trial-2.evxtOKBrowse... and navigate to C://Users/Reader Lab/2-colour-learning-experiment-2/ and name the file trial-x where x is
the number of the trial for that day (this can be found on the data sheet).Arena Settings tab and select Arena Settings 1Background.Open.Grab and make sure the arena graphics still line up then click Validate Setup.Trial ListAdd Trials in the top left corner and add 16 trialsAcquisition (0 acquired)Playback Control to set the background image.... next to Video under the Settings tab to the rightAnalysis tab go to the Results tab and click Statistics & ChartsCalculateExport tab click StatisticsOK, the default settings should be sufficientC://Users/Reader Lab/2-colour-learning-experiment-2/trial-x/Export Files/experiment-2-raw-data/Statistics-trial-x into the Google drive folder.